Abstract
CD8+ T cell immune responses are critical for combating infectious diseases and tumours1,2,3. Antigen cross-presentation, primarily occurring at the endoplasmic reticulum (ER) of dendritic cells, is essential for protein-based vaccines to induce CD8+ T cell responses4. Current efforts have focused on antigen delivery at the tissue and cellular levels, whereas subcellular delivery has been limited to facilitating antigen escape from lysosomes into the cytosol. In the absence of a small-sized high-affinity ER-targeting molecule, the importance of the ‘last mile’ from the cytosol to the ER remains elusive. Here we developed stimulator of interferon genes (STING) agonist-based ER-targeting molecules (SABER), which effectively deliver antigens to the ER and cluster key machinery in cross-presentation to form microreactors by folding the ER membrane. Conjugation of SABER to various antigens substantially enhances the induction of CD8+ T cell immune responses to tumour neoantigens and conserved viral epitopes, far exceeding that achieved by mixtures of antigens with STING agonists or conventional adjuvants. SABER also retains a potent adjuvant effect, effectively enhancing the ability of a SARS-CoV-2 subunit vaccine to induce broadly neutralizing antibodies. This study provides a high-affinity ER-targeting delivery system and vaccine adjuvant, demonstrating that precise subcellular delivery targeting the last mile of cross-presentation can lead to a qualitative leap.
Similar content being viewed by others
Main
Inducing robust CD8+ T cell immune responses is an important goal pursued by both cancer immunotherapies and infectious disease vaccines1,2,3. Dendritic cells (DCs) are key antigen-presenting cells (APCs) that prime CD8+ T cell responses by presenting antigens via major histocompatibility complex class I (MHC-I) and providing co-stimulatory signals5. Although significant progress has been made in enhancing co-stimulatory signals, improving antigen presentation remains essential for optimizing CD8+ T cell responses2,6,7.
Extracellular antigens, such as those found in peptide and subunit vaccines, require a specialized and less efficient process, known as cross-presentation, to be presented by MHC-I. The cytosolic pathway is the primary route for cross-presentation, involving antigen delivery to the cytosol and subsequent transport into the ER where peptides are further trimmed and loaded onto MHC-I4. Although previous efforts have largely focused on antigen uptake and cytosolic delivery, the last mile of ER targeting remains underexplored8. A few studies have explored ER targeting using signal sequences from adenovirus E3, but this long hydrophobic viral peptide may have limitations in clinical applications and its precise mechanism of action remains elusive9,10,11. A subsequent attempt using a shorter peptide did not achieve favourable targeting, highlighting the need for a high-affinity, small-sized ER-targeting molecule12.
Design of SABER
We first search for proteins residing on the outer surface of the ER that possess existing high-affinity small-molecule inhibitors or agonists. The STING protein emerged as one of the few candidates13. Natural agonists, such as cGAMP14 and c-diGMP15, and synthetic agonists, including diABZI16 and SR717 (ref. 17), all exhibit high affinity for STING, making them potential lead compounds.
An initial analysis of the chemical properties and crystal structures of STING–agonist complexes revealed that most agonists, including cGAMP and SR717, may not be suitable candidates. They induce a ‘closed lid’ conformation of STING, which wraps around the agonist and obscures potential antigen-conjugating sites17,18 (Extended Data Fig. 1a,b). c-diGMP and diABZI, a class of diamidobenzimidazole compounds, induce an ‘open lid’ conformation15,16 (Extended Data Fig. 1c,d). However, the optimal modification sites for c-diGMP, the 2-hydroxyl group and phosphates, are hidden at the bottom of the binding pocket (Extended Data Fig. 1c).
Our focus shifted to compound 2 in the diABZI family, whose 7-position of the benzimidazole is exposed and does not interact with STING, making it suitable for further modification16 (Extended Data Fig. 1d,e). We synthesized a series of compound 2 derivatives, and compared their STING activation capability with compound 3 (designated as diABZI hereafter), which was the most potent agonist within the diABZI family (Fig. 1a and Extended Data Fig. 1f). All molecules were validated by mass spectrometry and nuclear magnetic resonance. All chemical data can be found in Supplementary Information.
a, Chemical structures of the SABER family and diABZI. The EC50 was determined by measuring IFNβ in BMDCs for 24 h by ELISA. b, BMDCs from WT or STING−/− mice were incubated with 1 μM SABERs or diABZI for 4 h. Ifnb1 was measured by quantitative PCR with reverse transcription (RT–qPCR). n = 3 biologically independent samples. c, BMDCs were incubated with ABM5–OVA or SR717–OVA at different doses for 24 h, and measured for IFNβ. n = 2 biologically independent samples. d, WT or STING−/− BMDCs were incubated with 1 μM ABM5–OVA for 4 h, and measured for Ifnb1. n = 3 biologically independent samples. e, WT or STING−/− BMDCs were incubated with 0.5 μM OVA250–264 peptide, a mixture of diABZI and OVA peptide (diABZI + OVA) or ABM5–OVA for 8 h. f–h, The H2-Kb(SIINFEKL) complex (f), CD86 (g) and total H2-Kb (h) were analysed by flow cytometry. The MFI was summarized in f–h. n = 3 biologically independent samples. i,j, Cross-presentation (i) and CD86 (j) were monitored at 2, 4 and 8 h after treating BMDCs with 0.5 μM diABZI + OVA or ABM5–OVA. n = 3 biologically independent samples. k,l, WT or STING−/− BMDCs were treated with 50 nM OVA peptide, diABZI + OVA or ABM5–OVA for 8 h, and then co-cultured with CFSE-labelled OT-I cells for 72 h. The expansion of OT-I cells was measured by flow cytometry. The total number (k) and percentage (l) of expanded OT-I cells are summarized. n = 3 biologically independent samples. m,n, OT-I expansion was similarly measured in BMDCs receiving 50 nM OVA-i peptide, diABZI + OVA-i or ABM5–OVA-i. n = 3 biologically independent samples. Data are mean ± s.e.m. One-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test (f–h,k–n) and two-way ANOVA (i,j) were used. Data are representative of three independent experiments.
Half-maximal effective concentration (EC50) of each compound was determined using bone marrow-derived dendritic cells (BMDCs). ABM1, generated by introducing a maleimide-containing side chain at the 7-position, hardly activated STING (Fig. 1a). Replacing the single bond at the R1 position with a double bond (ABM2) or introducing a piperazine-containing side chain (ABM3) resulted in substantially enhanced potency. A combination of the double bond and piperazine-containing side chain (ABM5) further enhanced potency. In addition, the introduction of a longer side chain (ABM4) only slightly affected potency, providing another option for antigen conjugation.
Beyond maleimide, two derivatives (ABN1 and ABN2) with an azide group were synthesized, enabling conjugation with various unnatural amino acids. Both derivatives exhibited potency comparable with diABZI (Fig. 1a). This class of molecules was designed as SABER. We further confirmed that the efficacy of SABER depended on STING by using STING−/− BMDCs (Fig. 1b).
SABER promotes cross-presentation
ABM5 was selected as a representative to evaluate its capability for the cross-presentation of the MHC-I epitope OVA257–264 (SIINFEKL) of ovalbumin (OVA)2. ABM5 was conjugated to the N terminus of OVA250–264 (ABM5–OVA). This long peptide model antigen simulated the clinical scenario of tumour neoantigen vaccines, which typically contain 15–30 amino acids and require further trimming during cross-presentation5. Moreover, the long peptide containing this region was reported to effectively access the cytosolic pathway of cross-presentation, avoiding the interference of transfection reagents19. High-performance liquid chromatography and mass spectrometry revealed that ABM5 was successfully conjugated with the peptide, and the N-terminal cysteine reacted with maleimide to further cyclize, forming a more stable coupling20 (Extended Data Fig. 2a–c). Despite being linked to a long peptide, ABM5 retained its ability to effectively activate STING, albeit with slightly impaired EC50 (Fig. 1c and Extended Data Fig. 2d). By contrast, conjugating SR717 with the OVA peptide could not activate BMDCs (Fig. 1c). ABM5–OVA-mediated phosphorylation of STING and TANK-binding kinase 1 (TBK1) occurred at a slower pace but persisted longer than diABZI, resulting in slower degradation of STING (Extended Data Fig. 2e). Activity of ABM5–OVA fully depended on STING (Fig. 1d).
We next explored whether the OVA peptide, henceforth referred to as OVA, could be cross-presented more effectively following ABM5 decoration. Compared with OVA alone or an equimolar mixture of diABZI and OVA (diABZI + OVA), cross-presentation of SIINFEKL was significantly enhanced by ABM5–OVA (Fig. 1e). The mean fluorescence intensity (MFI) of the H2-Kb(SIINFEKL) complex was three times higher in the ABM5–OVA-treated group than in the mixture (Fig. 1f). Meanwhile, upregulation of the co-stimulatory factor CD86, a consequence of STING activation, was elevated in both diABZI + OVA and ABM5–OVA-treated groups (Fig. 1g). We further confirmed that enhanced cross-presentation depended on STING and increased peptide loading rather than an overall upregulation of H2-Kb (Fig. 1f,h).
Cross-presentation commenced as early as 2 h post-addition of ABM5–OVA and rapidly reached a significant level at 4 h, but was greatly slower in cells receiving diABZI + OVA (Fig. 1i). Conversely, upregulation of CD86 was slower in ABM5–OVA, consistent with its slower STING activation kinetics (Fig. 1j and Extended Data Fig. 2e). We subsequently confirmed that treating BMDCs with a tenfold lower concentration of ABM5–OVA maintained effective cross-presentation when diABZI + OVA was entirely ineffective (Extended Data Fig. 2f). The superior capability of SABER in enhancing cross-presentation was further confirmed by the OT-I assay. Again, the capability of ABM5–OVA depended on STING (Fig. 1k,l).
A mutated OVA250–264 peptide (L264I) with weaker immunogenicity, designated as OVA-i, was conjugated with ABM5 (ref. 21). ABM5–OVA-i also achieved significantly higher cross-presentation than diABZI + OVA-i, whereas retained capability in elevating CD86 levels (Fig. 1m,n and Extended Data Fig. 2g,h). These findings demonstrate that SABER decoration can enhance cross-presentation of both strong and weak epitopes.
ER targeting is the key mechanism
We next investigated whether enhanced cross-presentation was attributed to ER targeting. A new conjugate, ABD–S–OVA, was synthesized by incorporating a disulfide bond between the STING agonist and peptide (Extended Data Fig. 3a). This conjugate was designed to be rapidly cleaved by glutathione (GSH) within the cytosol, preventing peptides from being targeted to the ER22 (Extended Data Fig. 3b). As designed, ABD–S–OVA was quickly cleaved in 1 mM or 10 mM GSH (Extended Data Fig. 3c). Of note, multiple cleavage products were generated, reflecting a dynamic equilibrium between the primary cleavage product ABD–SH and GSH (Extended Data Fig. 3d–h). Nevertheless, ABD–S–OVA could effectively activate BMDCs, indicating that cleavage products retain biological activity (Extended Data Fig. 3b).
The subcellular distribution of Cy5.5 fluorophore-conjugated peptides was visualized. As previously reported in DCs19, we first confirmed that both OVA peptides and ABM5–OVA could effectively enter HeLa cells, whereas SNT, a peptide from SARS-CoV-2 (ref. 23), could not (Extended Data Fig. 4a,b). The activation of STING by ABM5–OVA suggested its translocation to the cytosol and ER, validating subsequently by immunofluorescence (Fig. 2a and Extended Data Fig. 4c). STING formed microclusters or puncta in diABZI or ABM5–OVA-treated cells, consistent with previous reports indicating that agonists induce STING aggregation24,25,26,27 (Fig. 2a). Only under the guidance of ABM5 did OVA peptides colocalize with STING microclusters, whereas diABZI + OVA or ABD–S–OVA only induced aggregation of STING but not peptides (Fig. 2a,b). Microclusters of peptides were not observed in STING−/− cells receiving ABM5–OVA (Fig. 2a). Furthermore, for peptides exhibiting limited cytosolic entry, such as ABM5–SNT, encapsulating them within lipid nanoparticles (LNPs) to facilitate cytosolic delivery also resulted in effective STING targeting28 (Extended Data Fig. 4d,e).
a,b, STING−/− HeLa cells and STING−/− HeLa cells expressing STING–Flag were incubated with 2.5 μM of Cy5.5-labelled OVA250–264 peptide (OVA–Cy5.5), diABZI + OVA–Cy5.5 or ABM5–OVA–Cy5.5 for 2 h, stained and analysed by confocal microscopy (a). Scale bars, 5 μm. The colocalization of STING and OVA peptides was calculated and is summarized (b). n = 20 cells. c,d, Cells were similarly treated as in panel a. The ER and Golgi were stained for calreticulin or GM130, respectively (c). Scale bars, 5 μm. The colocalization of peptide–ER and peptide–Golgi is summarized (d). n = 10 cells. e,f, HeLa cells expressing STING–Flag were incubated with 0.5 μM OVA–biotin peptide, diABZI + OVA–biotin, ABD–S–OVA–biotin or ABM5–OVA-biotin for 2 h. The ER was isolated by density gradient ultracentrifugation. OVA–biotin and calreticulin in the ER fraction were analysed by immunodot blot (e). For gel source data, see Supplementary Fig. 1a. The relative density of OVA–biotin and calreticulin is summarized (f). n = 3 biologically independent samples. g,h, The MFI of the H2-Kb(SIINFEKL) complex (g) and CD86 (h) were measured after treating BMDCs with 0.5 μM of ABD–S–OVA or ABM5–OVA. n = 3 biologically independent samples. i,j, BMDCs were treated with 50 nM ABD–S–OVA or ABM5–OVA for 8 h, and then co-cultured with OT-I cells for 72 h. The total number (i) and percentage (j) of expanded OT-I cells are summarized. n = 3 biologically independent samples. k, FLT3L-cultured BMDCs were incubated with 0.5 μM of OVA peptide, ABD–S–OVA, diABZI + OVA or ABM5–OVA for 8 h, and analysed for cross-presentation in cDC1 (CD11c+B220−CD24+CD172α−) and cDC2 (CD11c+B220−CD24−CD172α+). n = 3 biologically independent samples. Data are mean ± s.e.m. For the violin plot (e), the lines represent the median. One-way ANOVA with Tukey’s multiple comparisons test (b,d,f–j) and two-way ANOVA (k) were used. Data are representative of three independent experiments.
diABZI is known to induce rapid STING aggregation and translocation to the Golgi. Of note, the morphology of STING aggregates following ABM5–OVA treatment differed from those induced by diABZI or ABD–S–OVA, suggesting distinct subcellular localization (Fig. 2a). As shown in Fig. 2c,d, ABM5–OVA exhibited pronounced colocalization with ER but not Golgi, a pattern absent in STING−/− cells. To trace peptides in the ER, biotinylated OVA peptides, ABM5–OVA or ABD–S–OVA was used. ER isolation following various treatments revealed the highest level of OVA–biotin in cells receiving ABM5–OVA (Fig. 2e,f). These findings collectively indicate that ABM5 effectively targets antigens to the ER via STING binding.
To further corroborate the role of ER targeting on cross-presentation, BMDCs were pre-treated with diABZI to saturate STING and induce its translocation. As shown in Extended Data Fig. 4f,g, diABZI pre-treatment significantly impaired the ability of ABM5 to enhance cross-presentation. Conversely, pre-treatment with imiquimod, a TLR7 agonist, had no such effect. Moreover, although ABM5–OVA and ABD–S–OVA exhibited similar potency in upregulating CD86, ABD–S–OVA demonstrated a significantly reduced ability to enhance cross-presentation (Fig. 2g–j). This diminished efficacy was replicated when using the weaker OVA-i epitope (Extended Data Fig. 4h,i).
We further traced the intracellular dynamics of Cy5.5-labelled peptides. Consistent with previously observed activation kinetics, diABZI induced rapid STING aggregation followed by degradation at 8 h without promoting peptide clustering (Extended Data Figs. 2e and 4j). By contrast, ABM5–OVA and STING formed microclusters at 2 h and 4 h (Fig. 2a and Extended Data Fig. 4j). However, although STING persisted as microclusters, the Cy5.5 signal diffused away from STING at 8 h, suggesting that ABM5–OVA might be gradually cleaved upon reaching STING (Extended Data Fig. 4j,k).
Antigens were known to be processed by proteasomes or calpain in the cytosolic pathway before entering the ER lumen, or by lysosomal cathepsin S in the vacuolar pathway4. Cross-presentation of ABM5–OVA was substantially inhibited by MG132 but not by PD150606, indicating that the process follows a cytosolic pathway, with proteasomes mediating the cleavage (Extended Data Fig. 4l,m). Of note, a substantial portion of cytosolic proteasomes is tethered to the ER for ER-associated degradation29. OVA–ABM5, made by conjugating ABM5 to the C terminus of the peptide, exhibited significantly inferior capability in enhancing cross-presentation than ABM5–OVA (Extended Data Fig. 4n). Cy5.5-labelled OVA–ABM5 could colocalize with STING for a longer time, indicating that SABER modification at the C terminus may hinder proteasomal cleavage (Extended Data Fig. 4j,k). Inhibitors for ER aminopeptidase 1 (ERAP1) and MHC-I translocation further demonstrated that peptides must enter the ER for further trimming before being presented by MHC-I4 (Extended Data Fig. 4l,m). H151, an inhibitor that blocks palmitoylation of STING and subsequent TBK1 signalling in the Golgi, fully inhibited CD86 upregulation but only partially impaired cross-presentation, suggesting that STING binding rather than downstream signals contribute mainly to SABER-mediated cross-presentation enhancement30 (Extended Data Fig. 4m,o).
ABM5–OVA also exhibited significantly higher capability in enhancing cross-presentation in both conventional type 1 DC (cDC1) and cDC2, generated from FMS-like tyrosine kinase 3 ligand (FLT3L) culturing, than diABZI + OVA or ABD–S–OVA31 (Fig. 2k).
SABER condenses presentation machinery
The intracellular distribution of key machinery involved in the cytosolic pathway of cross-presentation, including STING, proteasome and the transporter associated with antigen processing (TAP), was next investigated. An enrichment of proteasomes and TAP1 was observed within STING microclusters in cells receiving ABM5–OVA but not diABZI + OVA (Extended Data Fig. 5a). We named this SABER-induced condensation of cross-presentation machinery as ‘microreactor’, which concentrates key machinery for cross-presentation within a confined space, accelerating the cleavage and transport of antigen peptides (Extended Data Fig. 5b). Formation of microreactors was subsequently confirmed by the STING fused with engineered ascorbic acid peroxidase 2 (STING–APEX2) system, which biotinylates proteins near STING32. This approach revealed a significant enrichment of TAP1 and multiple proteasome subunits close to STING following ABM5–OVA treatment compared with diABZI treatment (Extended Data Fig. 5c).
Furthermore, we demonstrated that treating BMDCs with SABER also clustered peptides with the ER, STING, proteasomes and TAP1 (Extended Data Fig. 5d–i). The formation of microreactors might be attributed to the folding of ER membranes caused by STING aggregation, which in turn caused agglomeration of proteasomes and TAPs attached to the ER4,26,29. Therefore, transmission electron microscopy was next used to visualize cellular ultrastructure. ABM5–OVA indeed promoted the clustering of smooth ER membranes around the nucleus, whereas diABZI induced a rapid increase in swelling the Golgi, probably due to the proton channel function of STING33 (Extended Data Fig. 5j). These findings demonstrate that SABER–peptide conjugates rather than diABZI effectively condense cross-presentation machinery into microreactors. The discrepancy between the two may be attributed to the slower kinetics of STING activation by conjugates, granting sufficient time for STING aggregation on the ER rather than being rapidly transported to the Golgi, a mechanism meriting further investigation.
SABER augments CD8+ T cell responses
The in vivo efficacy of SABER in augmenting CD8+ T cell immune responses was next explored. SABER–peptide conjugates were first encapsulated into LNPs containing the ionizable lipids SM102 or ALC0315 by microfluidic mixing to protect them and further facilitate their cellular entry28 (Extended Data Fig. 6a). LNPs containing 10% ALC0315 significantly augmented cross-presentation by more than twofold with acceptable particle size and polydispersity index, and were selected for subsequent studies (Extended Data Fig. 6b,c). Because encapsulating diABZI and OVA at a 1:1 molar ratio into the same LNP is technically challenging, we prepared LNPs encapsulating diABZI or OVA as controls. LNPs used in this study were about 100 nm in size with a polydispersity index below 0.2, and stable when being stored at 4 °C for 140 days (Extended Data Fig. 6d–g).
Encapsulating diABZI, ABM5–OVA or ABD–S–OVA into LNPs greatly improved their EC50 (Extended Data Fig. 6h). Besides mouse DCs, ABM5–OVA also effectively activated human monocyte THP-1-derived DC-like cells. Of note, encapsulating ABM5–OVA in LNPs further enhanced its efficacy (Extended Data Fig. 6i). BMDCs treated with LNP-encapsulated ABM5–OVA exhibited significantly enhanced cross-presentation compared with cells treated with a mixture of LNP-encapsulated diABZI and OVA, consistent with the findings from LNP-free cells (Fig. 3a). An OT-I assay confirmed that encapsulating ABM5–OVA into LNPs further enhanced cross-presentation (Fig. 3b,c). These results indicated that, although OVA peptides can enter the cytosol to a certain extent via an unknown mechanism, LNPs significantly enhanced this process19. Furthermore, LNPs should be crucial for most peptides that could not efficiently enter the cytosol and were used in subsequent in vivo studies.
a, BMDCs were incubated with 100 nM OVA peptide, diABZI + OVA or ABM5–OVA either carrier free or encapsulated in LNPs for 4 h. n = 3 biologically independent samples. b,c, BMDCs receiving 50 nM OVA peptide, ABM5–OVA or LNP-encapsulated ABM5–OVA for 4 h and then co-cultured with OT-I cells for 72 h. n = 3 biologically independent samples. All ABM5–OVA, ABD–S–OVA, OVA peptides and diABZI were encapsulated in LNPs and used at 10 nmol per mouse hereafter. d, Mice were immunized with diABZI + OVA or ABM5–OVA on days 0 and 14. OVA-tetramer+ CD8+ T cells in PBMCs were analysed on day 21. From left to right, n = 4, 4 and 6 mice. e, WT or STING−/− mice were immunized and analysed as in panel d. n = 4 mice. f,g, Mice were immunized with ABM5–OVA, or OVA peptide with or without 10 μg of ODN1018, ISCOMs or poly-I:C on days 0, 14 (f) and 28 (g). n = 5 mice. h, Mice were immunized with ABM5–OVA, diABZI + OVA or ABD–S–OVA for three doses. n = 5 mice. i, WT or Batf3−/− mice were immunized as in panel d. n = 4 mice. j,k, Mice were immunized as in panel d, challenged with 2 × 105 B16-OVA on day 28. From the top, n = 7, 8 and 8 mice. l–o, Mice were inoculated with 2 × 105 B16-OVA (l,m) or 5 × 105 E.G7-OVA (n,o), and received diABZI + OVA or ABM5–OVA on days 4, 11 and 18. B16-OVA. From the top, n = 8, 8 and 9 mice. E.G7-OVA n = 12 mice. p, Mice were immunized as in panel d, and challenged intravenously by 2 × 105 B16-OVA on day 28. Metastatic foci in the lungs were counted on day 46. n = 5 mice. Data are mean ± s.e.m. One-way ANOVA with Tukey’s multiple comparisons test (a–i,p), two-way ANOVA (j,l,n) and the log-rank test (k,m,o) were used. Data are representative of two independent experiments.
To be concise, ABM5–OVA encapsulated in LNPs are referred to as ABM5–OVA hereafter. The same applies to diABZI, peptides and other SABER–peptide conjugates unless otherwise specified. Subcutaneous administration of diABZI or ABM5–OVA into C57BL/6 mice did not induce significant local reactions, persistent and severe systemic side effects as assessed by body weight, blood chemistry, and serum interferon-β (IFNβ) and tumour necrosis factor (TNF) levels (Extended Data Fig. 6j–q). After two injections, ABM5–OVA conferred a ninefold increase in OVA-tetramer+ CD8+ T cells compared with diABZI + OVA (Fig. 3d). Their in vivo efficacy fully depended on STING (Fig. 3e). In addition to ALC0315, ABM5–OVA encapsulated in SM102-containing LNPs also induced a 17-fold enhancement compared with the mixture (Extended Data Fig. 7a).
We optimized the procedure to co-encapsulate diABZI and OVA within the same LNPs at a 1.2:1 ratio, which was still less effective than ABM5–OVA, excluding the possibility that the inferior efficacy of diABZI + OVA resulted from separate encapsulation of two components (Extended Data Fig. 7b). To exclude the possibility that the attenuated adjuvant effect was due to rapid release from LNPs, we replaced diABZI with a conjugate of ABM5 and a peptide from mutated ADP-dependent glucokinase (Adpgk), as an additional control. As shown in Extended Data Fig. 7c, ABM5–OVA remained significantly more effective than ABM5–Adpgk + OVA. We further demonstrated the necessity of LNPs for the in vivo application of SABER, although ABM5–OVA alone could induce a modest CD8+ T cell response (Extended Data Fig. 7c).
The efficacy of SABER was further compared with established adjuvants, including poly-I:C, ODN1018 and immune-stimulating complexes (ISCOMs). Both poly-I:C and ODN1018 demonstrated robust adjuvant activity, augmenting CD8+ T cell responses by ninefold and sevenfold over LNP-encapsulated OVA peptides alone, respectively. By contrast, ISCOMs exhibited limited efficacy as a standalone adjuvant in enhancing CD8+ T cell responses within this study (Fig. 3f). Of note, ABM5–OVA significantly outperformed these adjuvants, boosting CD8+ T cell responses by 58-fold over OVA alone (Fig. 3f).
Given the necessity of multiple booster vaccinations for emerging pathogens and cancer immunotherapies, we investigated the efficacy of SABER following a third dose. Although poly-I:C and ODN1018 induced only modest increases in CD8+ T cell responses beyond the second dose, ABM5–OVA significantly amplified immune responses, resulting in a 120-fold elevation compared with OVA alone (Fig. 3g). Of note, ABM5–OVA also demonstrated superior efficacy than ABD–S–OVA in vivo, indicating that the enhanced immune response is not solely attributable to the adjuvant effect but is dependent on ER targeting (Fig. 3h).
SABER mobilizes cDC1 and cDC2 in vivo
Next, we sought to identify DC subtypes involved in vivo. DiD-labelled ABM5–OVA-encapsulated LNPs primarily accumulated in the liver, spleen and draining lymph nodes (24 h after subcutaneous injection; Extended Data Fig. 7d,e). In draining lymph nodes, resident CD8+ cDC1s, migratory CD103+ cDC1s and CD11b+ cDC2s all captured LNPs (Extended Data Fig. 7f). Consistent with in vitro results, SABER significantly enhanced cross-presentation in both cDC1s and cDC2s, including CD103+ cDC1s, which are intrinsically proficient at cross-presentation (Fig. 2k and Extended Data Fig. 7g). Of note, diABZI + OVA did not enhance cross-presentation, despite both diABZI + OVA and ABM5–OVA promoting DC maturation (Extended Data Fig. 7g,h). In Batf3−/− mice that lack cDC1s, induction of CD8+ T cell responses by diABZI + OVA was almost completely ablated31 (Fig. 3i). By contrast, although the efficacy of ABM5–OVA was reduced in Batf3−/− mice, a significant level of CD8+ T cell response persisted, indicating that SABER could mobilize APCs other than cDC1s, such as cDC2s, for efficient cross-presentation (Fig. 3i).
SABER boosts antitumour immune responses
Immunized mice were next challenged with mouse melanoma B16F10-expressing OVA (B16-OVA). Coinciding with the highest level of OVA-specific CD8+ T cells, ABM5–OVA exhibited remarkable efficacy in inhibiting tumour growth, preventing all mice from succumbing to the disease for 5 weeks. By contrast, vaccination with diABZI + OVA only protected approximately 10% of mice (Fig. 3j,k).
ABM5–OVA also demonstrated efficacy as a therapeutic vaccine. Therapeutic injection of diABZI + OVA provided only limited suppression of tumour growth, and all mice succumbed within 5 weeks (Fig. 3l,m). Encouragingly, ABM5–OVA could provide prolonged tumour inhibition and prevent 75% of mice from death (Fig. 3l,m). We also confirmed the superior efficacy of ABM5–OVA in E.G7-OVA, a malignant T cell lymphoma expressing OVA (Fig. 3n,o). A metastasis model was further used by challenging immunized mice with B16-OVA intravenously. In contrast to mice receiving diABZI + OVA, mice receiving ABM5–OVA exhibited almost no visible tumour foci in the lungs (Fig. 3p).
To assess the applicability of SABER to longer peptides, we extended the OVA peptide by 15 amino acids at the C terminus, generating a 30-mer (OVA30aa) peptide (Extended Data Fig. 7i). Of note, ABM5–OVA30aa elicited CD8+ T cell responses sevenfold stronger than those induced by diABZI + OVA30aa (Extended Data Fig. 7j). The elevated immune response effectively rejected tumour growth (Extended Data Fig. 7k,l). This finding underscores the versatility of SABER for longer peptides with different epitope locations.
SABER potentiates neoantigen vaccines
We next investigated the effectiveness of SABER for bona fide tumour antigens. Adpgk has been identified as an MHC-I-restricted immunogenic neoantigen of mouse colorectal tumour MC38 (ref. 34). LNP-encapsulated ABM5–Adpgk neoantigen vaccine induced potent CD8+ T cell responses, which were sevenfold higher than those induced by diABZI + Adpgk (Fig. 4a and Extended Data Fig. 8a). As a result, three doses of ABM5–Adpgk were able to completely cure MC38 tumour-bearing mice (Fig. 4b,c). No tumour was detected until 90 days. By contrast, diABZI + Adpgk showed no significant benefit (Fig. 4b,c). The immune response induced by ABM5–Adpgk was long-lasting, as tumours could no longer be successfully seeded by a second challenge on day 90 (Fig. 4d and Extended Data Fig. 8b). Adpgk-specific CD8+ T cell responses were still detectable 60 days later (Fig. 4e). More than 80% of these cells were central memory T cells (Extended Data Fig. 8c).
a, Mice received 10 nmol ABM5–Adpgk or diABZI + Adpgk for three doses, and were measured for Adpgk-tetramer+ CD8+ T cells. From left to right, n = 7, 7 and 9 mice. b,c, Mice were inoculated with 1 × 105 MC38 and received 10 nmol ABM5–Adpgk or diABZI + Adpgk on days 4, 11 and 18. From the top, n = 13, 16 and 18 mice. d, Survived mice were rechallenged on day 90. n = 9 mice. e, Splenic tetramer+ CD8+ T cells were analysed 60 days after rechallenge. From left to right, n = 6 and 8 mice. f,g, Mice were inoculated with 1.5 × 105 B16F10 and received 10 nmol ABN2–M27, diABZI + M27 and poly-I:C (10 μg) + M27 on days 6, 13 and 20, with or without 200 μg anti-PD1. n = 5 mice. h,i, Mice received 30 nmol ABM5–SNT or diABZI + SNT for two doses. ELISpot and intracellular cytokine staining were used to measure IFNγ-expressing cells restimulated with SNT. n = 6 mice. j,k, hACE2-Tg mice were similarly immunized, challenged with SARS-CoV-2 BA.5.2 on day 21 and analysed for viral loads. n = 5 mice. l, Mice received 100 μg OVA protein alone or with 10 nmol diABZI, ABM5–SNT, poly-I:C (10 μg) or ISCOMs (10 μg) on days 0 and 14. OVA-specific IgG was measured on day 21. n = 6 mice. m,n, Mice received 10 μg RBD-Fc with 10 nmol diABZI + SNT or ABM–SNT on days 0 and 14. WT RBD-specific IgGs were measured on days 14 and 21. n = 6 mice. Data are mean ± s.e.m. (a,b,d–f,h,i) and geometric mean ± 95% CI (j–n). One-way ANOVA with Tukey’s multiple comparisons test (a,h,i), two-tailed Student’s t-test (e,j,k), two-way ANOVA (b,d,f), log-rank test (c,g) and Kruskal–Wallis test (l–n) were used. Data are representative of two independent experiments.
We further investigated the efficacy of SABERs in wild-type (WT) B16F10, known for its resistance to immune checkpoint therapy. A neoantigen tumour vaccine based on an MHC-I epitope M27 was developed35. Of note, the presence of cysteine within this epitope necessitated a switch from ABM5 to ABN2 (Fig. 1a and Extended Data Fig. 8d). To mimic clinical scenarios, treatment was initiated at a later time point, 6 days post-tumour inoculation. ABN2–M27 alone significantly inhibited tumour growth compared with diABZI + M27 (Fig. 4f,g). Although anti-PD1 monotherapy exhibited limited therapeutic benefit against B16F10, its combination with ABN2–M27 demonstrated superior efficacy than all other treatment groups, including poly-I:C + M27 + anti-PD1 (Fig. 4f,g). Poly-I:C-based adjuvants and anti-PD1 are widely used in clinical studies of neoantigen vaccines36. These findings underscore the efficacy of SABER-based neoantigen vaccines in a challenging tumour model and highlight the adaptability of the platform to diverse peptide linkage requirements.
SABER boosts antiviral immune responses
Induction of CD8+ T cells against conserved epitopes is a critical objective of vaccines for rapidly evolving pathogens such as SARS-CoV-2. To evaluate the efficacy of SABER, we initially tested it on a peptide vaccine targeting N129–148 (SNT), which contains conserved T cell epitopes for both mice and humans23,37 (Extended Data Fig. 8e). As demonstrated by enzyme-linked immunosorbent spot assay (ELISpot), ABM5–SNT immunization generated a 150-fold higher number of IFNγ-expressing T cells over diABZI + SNT (Fig. 4h and Extended Data Fig. 8f). Intracellular cytokine staining further confirmed that ABM5–SNT was highly effective in inducing IFNγ-expressing CD8+ T cells against SNT (Fig. 4i and Extended Data Fig. 8g). The human angiotensin-converting enzyme 2 transgenic (hACE2-Tg) mice were immunized and challenged with the SARS-CoV-2 Omicron variant BA.5.2. Encouragingly, the potent immune responses against a single peptide induced by ABM5–SNT could reduce viral loads in both the lungs and brain by 100-fold (Fig. 4j,k).
We further investigated whether SABER–peptide conjugates could serve as an adjuvant to enhance antibody induction. When ABM5–SNT was co-immunized with OVA protein, it potently augmented antibody responses in primary and booster immunizations, exhibiting comparable or even superior adjuvant effects to diABZI, poly-I:C and ISCOMs (Fig. 4l and Extended Data Fig. 8h). To mimic the potential clinical application of ABM5–SNT, we co-immunized it with RBD-Fc, a subunit vaccine of SARS-CoV-2 (ref. 38) (Extended Data Fig. 8i). Although Delta variant-derived RBD-Fc alone barely induced antigen-specific IgGs against WT RBD even after booster, a single dose of RBD-Fc + ABM5–SNT was sufficient to induce high titres of RBD-specific IgGs, which could be further boosted by a second dose (Fig. 4m,n). Although RBD-Fc elicited substantial neutralizing antibody (NAb) titres to the Delta variant, its ability to induce cross-NAbs to mismatched viral strains, such as WT, BA.1 and BA.5, was considerably weaker (Extended Data Fig. 8j–m). Of note, ABM5–SNT exhibited a remarkable adjuvant effect in this scenario, boosting cross-NAbs against WT by over tenfold (Extended Data Fig. 8k). In the presence of ABM5–SNT, RBD-Fc could induce a significant but low level of cross-NAbs against BA.1 and BA.5, suggesting that induction of CD8+ T cells targeting conserved epitopes will be beneficial. Besides humoral immune responses, ABM5–SNT also retained adjuvant effects in augmenting T helper 1 CD4+ T cell responses induced by RBD-Fc (Extended Data Fig. 8n,o). These results demonstrated that a combination of SABER–peptide vaccines and subunit vaccines could induce potent antiviral CD8+ T cell immune responses and cross-NAbs, mobilizing both cellular and humoral arms of the immune system.
Personalized tumour vaccines have garnered significant attention as a promising approach to cancer treatment. Peptide vaccines derived from tumour-specific antigens represent an important category of such vaccines36,39,40. Similarly, peptide vaccines targeting conserved viral epitopes offer a potential strategy for developing universal influenza and pan-sarbecovirus vaccines41,42. However, low efficiency of cross-presentation often limits the induction of CD8+ T cells by these vaccines, leading to a predominance of CD4+ T cell responses1,5. Extensive efforts have been devoted over the past decades to enhance cross-presentation by delivering antigens to lymph nodes or DCs and facilitating lysosomal escape, such as LNPs utilized in this study43,44,45,46. Building on many pioneering studies on cross-presentation, STING and the ER, we were finally able to develop SABER, a family of high-affinity small-sized ER-targeting molecules2,9,10,11,12,13,14,15,16,17,18,24,25,26,27,30,32,33. By binding STING, SABER could target peptides to the ER, enrich key machinery required for cross-presentation and activate co-stimulatory signals (Extended Data Fig. 9). SABER is compatible with various peptide lengths, sequences and epitope positions, profoundly promoting the induction of CD8+ T cell immune responses by multiple tumour or viral peptide vaccines tested in this study. SABER–peptide conjugates are approximately 7–150 times more effective than a mixture of STING agonists and peptides in different vaccines and assays, demonstrating that ER targeting can lead to a substantial improvement.
SABER augments greatly higher CD8+ T cell responses than diABZI, poly-I:C, ODN1018 and ISCOMs, while maintaining a similar or even stronger adjuvant effect to promote humoral immunity. These benchmark adjuvants and their derivative are widely used in cancer and infectious disease vaccines36,39,40,47,48. Although neoantigen vaccines have been formulated with potent adjuvants such as poly-I:C, induction of CD8+ T cell responses has been suboptimal in both our mice models and clinical trials36,39,40. SABER could convert 30% of circulating CD8+ T cells into antigen-specific cells, a proportion that rarely exceeded 5% by other adjuvants used in this study for the same antigens. The robust efficacy of SABER in mice fuels our enthusiasm for preclinical models and clinical studies.
Methods
Synthesis of the SABER family
The interaction diagram of compound 2 and STING was analysed by Ligplot+ software (v2.2). Detailed synthetic procedures for the SABER family have been provided in the Supplementary Information. In brief, methyl-4-chloro-3-methoxy-5-nitrobenzene carboxylate was suspended in saturated ammonia and stirred at room temperature for 24 h, followed by heating at 50 °C for 2 h to obtain compound 1. Compound 1 was dissolved in dichloromethane (DCM), then boron bromide was injected dropwise, and stirred at room temperature for 16 h. The result solution was poured into ice water and stirred vigorously for 30 min to obtain compound 6. Compound 6 was dissolved in N,N-dimethylformamide (DMF). Then, tert-butyl 4-(3-(tolylenesulfonyloxy) propyl) piperazine-1-carboxylate and N,N-diisopropyl-ethylamine (DIPEA) were added and stirred at 80 °C for 4 h to obtain compound 7. The parallel synthesis process was initiated by dissolving compound 1 in DMF, followed by adding tert-butyl (4-aminobut-2-en-1-yl) carbamate and DIPEA and heating at 120 °C for 16 h. The crude solution was dissolved in DCM, and stirred with trifluoroacetic acid (TFA) to obtain compound 4. The dried compounds 4 and 7 were dissolved in isopropanol, and heated at 120 °C for 21 h with DIPEA. The crude solution was cooled. The brick-red precipitation was collected, dissolved in methanol with sodium dithionite, stirred at room temperature for 15 min, and quenched by adding sodium bicarbonate to obtain compound 15. Compound 15 was dissolved in DMF, followed by dropwise injection of 1-ethyl-3-methyl-1H-pyrazole-5-carbonyl isothiocyanate (in 1,4-dioxane) in the ice bath, and stirred at 0 °C for 30 min. Then, N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDCI) and DIPEA were added and stirred overnight at room temperature to obtain compound 20. Compound 20 was dissolved in DMF and stirred at room temperature after adding TFA for 2 h. The crude product was then dissolved in DMF. N-succinimidyl 6-maleimidohexanoate and DIPEA were added to the solution in the ice bath. The mixture was slowly warmed up to room temperature with stirring for 1 h to obtain ABM5. Compounds were purified by column chromatography, qualified and quantified by nuclear magnetic resonance (Avance III, Bruker) and mass spectrometry (Orbitrap Exploris 480, Thermo Fisher).
All peptides including OVA (CSGLEQLESIINFEKL), OVA-i (CSGLEQLESIINFEKI), OVA–biotin (CSGLEQLESIINFEKbiotinL), OVA30aa (CSGLEQLESIINFEKLTEWTSSNVMEERKIK), SNT (CKDGIIWVATEGALNTPKDHIGT), Adpgk (CELASMTNMELM) and M27 (propargylglycine-REGVELCPGNKYEM) were synthesized by Syneptide. To obtain SABER–peptide conjugates, ABM5 was covalently coupled with cysteine added to the N terminus of peptides. ABM5 in DMF was mixed with peptides in DMSO at a molar ratio of 1.1. Then, an equal volume of PBS was added. The mixture was shaken at room temperature for 24 h, followed by adding 10 times equal molar of DIPEA, and shaken at room temperature overnight. For the disulfide-containing counterpart, ABD in DMF was mixed with peptides in DMSO containing 4.8% acetic acid and incubated overnight at 37 °C. To covalently conjugate M27 with ABN2, ABN2 in DMSO was mixed with M27 in DMSO at a molar ratio of 2. Then, 8.8 times equal molar of CuSO4 (aqueous) and 17.6 times equal molar of sodium ascorbate (aqueous) were added, and shaken at room temperature for 1 h. To be labelled by Cy5.5, peptides or SABER–peptides in DMSO were mixed with Cy5.5–NHS in DMSO and DIPEA at molar ratios of 1 and 2, respectively. The mixture was shaken overnight at room temperature. All crude products obtained above were purified by high-performance liquid chromatography (HPLC; 1260, Agilent) and freeze-dried (FD-1A-50F+, Biocool). The molecular weight was confirmed by mass spectrometry.
To investigate the stability of conjugates, 10 μM ABM5–OVA or ABD–S–OVA were incubated with 1 mM or 10 mM GSH in PBS at 37 °C for 5 min, 1, 4 and 24 h, and then analysed by HPLC at 321 nm. The cleavage products were further confirmed by mass spectrometry.
Cell lines
STING−/− HeLa cells were provided by Z. Jiang at Peking University26. E.G7-OVA (CRL-2113) and Vero-E6 (CRL-1586) cells were obtained from the American Type Culture Collection. hACE2 stable expressing the HEK293T cell line (hACE2-293T) was obtained from PackGene Biotech. HeLa, THP-1, HEK293T, MC38 and B16F10 cell lines were maintained in our laboratory and validated by the STR identity. To generate B16-OVA cells, lentiviruses coding OVA were made by co-transfection of HEK293T cells with pCDH-ovalbumin-puro, psPAX2 and pMD2G plasmids. B16F10 cells were infected by the lentivirus and screened for single clones expressing OVA in a medium containing 2 μg ml−1 puromycin (MB2005, Meilunbio). The expression of OVA was validated by western blot. Only monoclonal cell lines that could express OVA over 15 generations were maintained and used for the following experiments. All cells were tested for Mycoplasma by a PCR kit (KP213-01, Tiangen).
Mice
C57BL/6 mice were obtained from GemPharmatech or the animal facility of Sun Yat-sen University. OT-I CD45.1+ transgenic mice were provided by C. Y. Yang at Sun Yat-sen University. STING−/− mice were provided by H. Yin at Tsinghua University, and were originally generated by Z. Jiang at Peking University. Batf3−/− mice were obtained from The Jackson Laboratory. Mice were housed in a specific pathogen-free environment in the animal facility of Sun Yat-sen University under 12-h light–dark cycles with ambient temperature between 21 and 24 °C and humidity between 40 and 70%. All the animal care and experimental procedures were performed with ethical compliance and approval by the IEC for Clinical Research and Animal Trials of the First Affiliated Hospital of Sun Yat-sen University or the Institutional Animal Care and Use Committee of Sun Yat-sen University. Male human ACE2-Tg (hACE2-Tg, CAG-hACE2-IRES-Luc-Tg) were purchased from the Shanghai Model Organisms Center, and housed in the animal facility of the School of Basic Medical Sciences at Fudan University. The protective effectiveness evaluation of vaccination in the mouse model of BA.5.2 infection was approved by the Laboratory Animal Ethics Committee of the School of Basic Medical Sciences at Fudan University.
Viruses
SARS-CoV-2 BA.5.2 variant was obtained from Shanghai Medical College, Fudan University, and the strain has been verified by next-generation sequencing twice. A plaque assay was used to quantify the viral titres using Vero-E6 cells. Experiments related to authentic SARS-CoV-2 were conducted in the BSL-3 Laboratory of Fudan University.
In vitro DC culture
Bone marrow cells were obtained from tibiaes and femurs of 4–6-week-old C57BL/6 mice. BMDCs were cultured as previously described7. In brief, bone marrow cells were cultured with 20 ng ml−1 of mouse granulocyte–macrophage colony-stimulating factor (GM-CSF; 576306, BioLegend) at the concentration of 1 × 106 per millilitre in complete RPMI-1640 (supplemented with 100 U ml−1 penicillin, 100 μg ml−1 streptomycin and 10% FBS) for 7 days at 37 °C and 5% CO2. FLT3L BMDCs were cultured as previously described49. In brief, bone marrow cells were resuspended at 2 × 106 per millilitre in complete RPMI-1640 containing 200 ng ml−1 recombinant FLT3L (550602, BioLegend) for 9 days at 37 °C and 5% CO2. THP-1-derived DC-like cells were cultured as previously described50. In brief, 2 × 105 per millilitre THP-1 cells were cultured in complete RPMI-1640 supplemented 0.05 mM 2-mercaptoethanol, 100 ng ml−1 human recombinant GM-CSF (572902, BioLegend) and 100 ng ml−1 human recombinant IL-4 (574002, BioLegend). The cells were cultured for 5 days at 37 °C in 5% CO2, with a media change on day 3.
EC50 measurement
The EC50 was determined by IFNβ expression in BMDCs. In brief, 1.5 × 104 BMDCs were seeded in 96-well plates and stimulated with SABERs, diABZI (tlrl-diabzi, InvivoGen), conjugates or their LNP-encapsulated counterparts at various concentrations. ABMs and ABNs were first blocked with cysteine or alkynylglycine plus ‘click’ reagents, respectively. After 24 h of inoculation, the supernatant of cells was collected for detecting IFNβ by an ELISA kit (439404, BioLegend). The EC50 was calculated from an agonist versus response curve by GraphPad Prism (v9.5.1).
Cytokine mRNA detection
Total RNA from BMDCs, HeLa, THP-1 or THP-1-derived DC-like cells was purified using the RNA Preparation Kit (AP-MN-MS-RNA-250, Axygen). cDNA synthesis was performed using the Thermo Scientific RevertAid RT Kit (K1691, Thermo Fisher). Real-time PCR was performed on the QuantStudio Real-Time PCR Detection System (Thermo Fisher) using the TB Green Master Mix (RR820Q, TaKaRa). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) served as an internal control. The primers for the detection of cytokine mRNA in BMDCs were: Cxcl10 forward: CCAAGTGCTGCCGTCATTTTC; Cxcl10 reverse: TCCCTATGGCCCTCATTCTCA; Gapdh forward: ATCAAGAAGGTGGTGAAGCA; Gapdh reverse: AGACAACCTGGTCCTCAGTGT; Ifnb1 forward: AGCTCCAAGAAAGGACGAACA; Ifnb1 reverse: GCCCTGTAGGTGAGGTTGAT.
The primers for detection of cytokine mRNA in HeLa, THP-1 or THP-1-derived DC-like cells were:
Cxcl10 forward: GTGGCATTCAAGGAGTACCTC; Cxcl10 reverse: TGATGGCCTTCGATTCTGGATT; Gapdh forward: GGAGCGAGATCCCTCCAAAAT; Gapdh reverse: GGCTGTTGTCATACTTCTCATGG; Ifnb1 forward: GCTTGGATTCCTACAAAGAAGCA; Ifnb1 reverse: ATAGATGGTCAATGCGGCGTC.
Immunofluorescence
For the colocalization of peptides, STING and various organelles, 1 × 105 HeLa or STING−/− HeLa cells were transfected with pCMV-STING1(human)-3×FLAG-WPRE-Neo in the presence of 10 μM cGAS inhibitor G140 (HY-133916, MCE) 24 h before the stimulation. BMDCs were seeded as mentioned above. The cells were treated with 5 μM biotin-labelled molecules, 2.5 μM Cy5.5-labelled molecules or 1 μM LNP-encapsulated Cy5.5-labelled molecules for the indicated time, fixed with 4% paraformaldehyde, and permeabilized by a permeabilization buffer containing saponin (P0095, Beyotime) with 30% FBS or normal goat serum. Samples were incubated with anti-FLAG (1:100; F1804, Sigma-Aldrich), anti-calreticulin (1:100; ab2907, Abcam), anti-TAP1 (1:20; 11114-1-AP, Proteintech), anti-proteasome 20S (1:100; ab22673, Abcam), anti-GM130 (1:100; 618022, BD) or anti-STING (1:100; MABF270, Sigma-Aldrich) antibodies at 4 °C overnight. Cells were washed and incubated with secondary antibodies at room temperature for 1 h, including AF488-conjugated anti-mouse IgG (1:300; 4408, CST), AF555-conjugated anti-rabbit IgG (1:300; ab150074, Abcam), STAR-ORANGE anti-mouse IgG (1:100; STORANGE-1001, Abberior) and STAR-RED anti-rabbit IgG (1:100; STRED-1002, Abberior). Cells were subsequently stained with DAPI (1:500; D9542, Sigma Aldrich) for 10 min at room temperature. The stained samples were mounted by ProLong Gold Antifade Mountant (P36982, Thermo Fisher) and subjected to imaging with confocal fluorescence microscopy (FV-3000, Olympus) or STED (STEDYCON, Abberior). FV31S-SW and STEDYCON Gallery software were used to collect and analyse imaging data. Mander’s colocalization coefficiency was calculated by JACoP plugin of ImageJ51.
TEM
To visualize cellular ultrastructure after SABER stimulation, 1 × 105 BMDCs were seeded in 12-well plates. Cells were incubated with 2.5 μM of diABZI or ABM5–OVA for 2 h, with PBS-treated cells serving as the negative control. Cells were fixed with 2.5% glutaraldehyde in PBS for 15 min at room temperature. Subsequently, the cells were stained with OsO4–potassium ferrocyanide, dehydrated and embedded in Epon. The ultrathin sections were obtained and examined by TEM (Talos L120C G2, Thermo Scientific).
Antigen cross-presentation and maturation of DCs
To evaluate antigen cross-presentation and STING activation-mediated upregulation of co-stimulatory factors, 4 × 105 BMDCs, STING−/− BMDCs or FLT3L BMDCs were pulsed by either peptides with or without diABZI or various conjugates for the indicated time. For inhibitor studies, various inhibitors, including MG132 (10 μM; HY-13259, MCE), PD150606 (10 μM; 26066, MCE), ERAP1-IN-1 (100 μM; HY-133125, MCE), diABZI (10 μM), H151 (10 μM; T5674, TargetMol) or imiquimod (10 μM; HY-B0180, MCE) were used to treat BMDCs 40 min before addition of ABM5–OVA, except brefeldin A (10 μg ml−1; 420601, BioLegend), which was added 2 h ahead. Cells were then pulsed with 0.5 μM ABM5–OVA for 8 h. BMDCs were then harvested and stained by FITC–anti-CD11c (1:100; 117306, BioLegend), APC–anti-H2-Kb-SIINFEKL (1:100; 141606, BioLegend), BV421–anti-CD86 (1:100; 105032, BioLegend), PE–anti-H2-Kb (1:100; 116508, BioLegend) and Live/Dead dye (1:400; 423107, BioLegend). In FLT3L BMDCs, cDC1s and cDC2s were identified by FITC–anti-CD11c (1:100; 117306, BioLegend), APC–anti-CD24 (1:100; 138505, BioLegend), BV421–anti-CD86 (1:100; 105032, BioLegend), PE/Cy7–anti-B220 (1:100; 103222, BioLegend), PerCP/Cy5.5–anti-CD172α (1:100; 144010, BioLegend), PE–anti-H2-Kb-SIINFEKL (1:100; 141603, BioLegend) and Live/Dead dye (1:400; 423107, BioLegend). The measurement of in vivo antigen cross-presentation and related DC subtypes was performed as in previous studies31,52. Mice receiving subcutatneous injection of 10 nmol LNP-encapsulated ABM5–OVA were euthanized 24 h later, and then draining lymph nodes were collected. Single cells were obtained by passing the tissues through 40-μm cell strainers, lysed with ACK lysis buffer (BL503A, Biosharp), blocked by anti-CD16/CD32 (1:400; 101301, BioLegend), and stained with FITC–anti-CD11c (1:100; 117306, BioLegend), BV711–anti-CD8a (1:100; 100747, BioLegend), APC/Cy7–anti-CD103 (1:100; 121431, BioLegend), PE–anti-CD11b (1:100; 101207, BioLegend), PB–anti-CD86 (1:100; 105021, BioLegend) and PE/Cy7–anti-H2-Kb-SIINFEKL (1:100; 141607, BioLegend). The stained samples were examined by flow cytometry (FACS LSRFortessa, BD). The data were collected using BD FACSDiva software and analysed by FlowJo (v10.8).
OT-I assay
BMDCs were first pulsed by either 50 nM peptides with or without diABZI or various conjugates for 8 h, or their LNP-encapsulated counterparts for 4 h, and then washed thoroughly before co-culturing with OT-I cells. CD8+ T cells were isolated from OT-I mice by the EasyStep Mouse CD8+ T Cells isolation Kit (19853A, StemCell), stained with CFSE Cell Division Tracker Kit (1:1,000; 423801, BioLegend), and incubated with BMDCs at a 20:1 ratio for 72 h. The data were collected using BD FACSDiva software and analysed by FlowJo (v10.8).
STING–APEX2 assay
HEK293T cells were first treated with the 10 μM cGAS inhibitor G140 1 h before transfection, and then transfected with pcDNA3.1-STING-APEX232. Twenty-four hours later, cells were stimulated with 0.5 μM diABZI, OVA or ABM5–OVA for 2 h. Before harvesting, cells were incubated with 500 μM biotin-phenol (HY-125658, MCE) in a complete medium for 30 min at 37 °C, followed by treatment with 1 mM H2O2 (AAPR43-100, Pythonbio) at room temperature for 2 min. Cells were then washed three times with fresh quencher solution containing 10 mM sodium ascorbate (V900326, Sigma-Aldrich) and 5 mM Trolox (HY-101445, MCE) in PBS. After washing, the cells were lysed in 600 μl of fresh RIPA lysis buffer (HY-K1001, MCE) supplemented with 1 mM PMSF (ST506, Beyotime), 5 mM Trolox and 10 mM sodium ascorbate, assisted by sonification. The lysate was then centrifuged at 13,400g for 20 min at 4 °C to pellet the cell debris. Protein concentration was measured using the BCA assay (P0012S, Beyotime). Clarified cell lysates were incubated with 50 μl of streptavidin magnetic beads (11641778001, Roche) at 4 °C overnight on a rotator to pull down biotinylated proteins. The beads were then washed twice with RIPA lysis buffer, once with 0.2 M KCl, once with 0.1 M Na2CO3, once with 2 M urea in 10 mM Tris-HCl (pH 8.0), and twice again with RIPA lysis buffer at 4 °C. Biotinylated proteins were eluted by boiling the beads in 75 μl of protein loading buffer supplemented with 0.67 mM biotin and 6.7 mM dithiothreitol for 10 min. The eluates were collected by pelleting the beads using DynaMag magnets (12321D, Invitrogen) and analysed by western blot.
Western blot
BMDCs receiving various treatments were lysed by RIPA. BMDC lysate in RIPA or final eluates from the STING–APEX2 assay were prepared in protein loading buffer and stored at −20 °C until use. The membranes were first stained by anti-STING (1:2,000; 13647, CST), anti-TBK1 (1:2,000; 3504T, CST), anti-pSTING (1:500; 72971S, CST), anti-pTBK1 (1:500; 5483T, CST), anti-TAP1 (1:1,000; 11114-1-AP, Proteintech), anti-Proteasome 20S (1:1,000; ab22673, Abcam) and anti-GAPDH (1:3,000; ABL1020, Abbkine). The corresponding horseradish peroxidase (HRP)-conjugated anti-mouse IgG (1:5,000; 7076, CST) or HRP-conjugated anti-rabbit IgG (1:5,000; 7074, CST) were used as secondary antibodies. After extensive washing, the bands were developed using an enhanced chemiluminescent detection kit (180-5001, Tanon) and visualized by an imaging system (ImageQuant 800, Cytiva).
ER isolation and immunodot blot
HeLa cells were transfected by pCMV-STING1(human)-3×FLAG-WPRE-Neo for 24 h, and pulsed by 0.5 μM of OVA–biotin, diABZI + OVA–biotin, ABD–S–OVA–biotin or ABM5–OVA–biotin for 2 h. ER was isolated by an ER isolation kit (ER0100, Sigma-Aldrich). The ER-containing fraction was collected and spotted on the polyvinylidene fluoride (PVDF) membrane. The film was blocked, washed and stained by streptavidin–HRP (1:1,000; 21130, Thermo Scientific), or anti-calreticulin (1:1,000; ab2907, Abcam), followed by washing and staining with HRP-conjugated anti-rabbit IgG (1:5,000; 7074, CST) as secondary antibodies. After extensive washing, the dots were developed using an enhanced chemiluminescent detection kit, visualized by an imaging system.
LNPs
LNPs were prepared by microfluidic mixing. In brief, all lipid materials were dissolved in ethanol at 10 mg ml−1 as stock solutions. The mass ratio of LNP was ionizable lipid SM102 (SN-M-CL23, SUNA Med-Engineering) or ALC0315 (SN-M-CL19, SUNA Med-Engineering):DSPC (LP-R4-076-1, Ruixibio):cholesterol (C3045, Sigma-Aldrich):DSPE-PEG2000 (880120P, Avanti Polar Lipids) at 10:50:37.5:2.5. Peptides, diABZI or ABM5–peptide conjugations were diluted in PBS at 333 μg ml−1 as aqueous phase. The mixture of lipids was used as the ethanol phase. For DiD-labelled LNPs, DiD was dissolved in the ethanol phase at the final concentration of 100 μg ml−1. The ethanol phase and aqueous phase were placed in a LNP packaging system (INano L, Micro & Nano) with a 1-ml or 2.5-ml syringe, squeezed at a total flow rate of 12 ml min−1 with the volume ratio ethanol:aqueous = 1:3. Fresh nanoparticles were then dialysed against PBS through 50 kDa MWCO dialysis bags (AFH0318, Milone) with gentle agitation at 4 °C for 6 h, changing the PBS every 2 h. The size of the LNPs was measured by Particle Analyzer (Litesizer 500, Anton Paar). Encapsulation efficiency was determined by a BCA Protein Assay Kit of LNP-containing solely peptides and compared with the standard curve of peptides under the same conditions. LNPs containing diABZI and SABER–peptide conjugates were quantified by HPLC. The resulting LNPs were stored at 4 °C.
Preparation of ISCOMs
The ISCOMs adjuvant used in this study is a self-assembled, saponin-based nanoparticle prepared as previously described53. In brief, cholesterol and phosphatidylcholine were dissolved in ethanol at the final concentration of 10 mg ml−1, respectively. Meanwhile, 4 mg Quil-A (vac-quil, InvivoGen) was dissolved in 3 ml of sterile PBS. Then, 80 μl of each cholesterol and phosphatidylcholine solution was mixed first, then injected into the Quil-A solution dropwise under the stirring condition. The mixture was allowed to stir for a further 2 h at room temperature, followed by dialysis against PBS using a 10 kDa MWCO dialysis bag (AFH0272, Milone) for 24 h at room temperature. After dialysis, the mixture was further purified by passing through the PD-10 desalting column (Sephadex G-25M, 17085101, GE). The pass-through solution was collected and determined the particle size by Particle Analyzer. Only the fraction with an average particle size of less than 50 nm was collected and used as the adjuvant. The ISCOMs adjuvant was quantified according to the content of Quil-A by HPLC.
Immunization and immune responses
C57BL/6 mice, 6–8 weeks old, were subcutaneously immunized with various free or LNP-encapsulated peptides, with or without free or LNP-encapsulated diABZI, free poly-I:C (tlrl-pic, InvivoGen), ODN1018 (HY-150724C, MCE) or ISCOMs, serving as controls or benchmarks. Peptides and diABZI were used at 10 nmol per mouse. Poly-I:C, ODN1018 and ISCOMs were used at 10 μg per mouse. Other mice were subcutaneously immunized with free or LNP-encapsulated SABER–peptide conjugates at 10 nmol per mouse. All mice were immunized on days 0 and 14, or received a third dose on day 28. Batf3−/− and STING−/− mice were similarly immunized, and WT C57BL/6 mice with the same age and gender were used as controls. The peripheral blood and spleens were collected 7 days after the last injection. The peripheral blood was lysed with ACK lysis buffer twice, and washed with PBS containing 1% FBS. Spleens were processed into single-cell suspensions by passing the tissues through 40-μm cell strainers, lysed with ACK lysis buffer, and the remaining cells were washed with PBS containing 1% FBS. To analyse the CD8+ T cell response, PBMCs and splenocytes were blocked by anti-CD16/CD32 (1:400; 101301, BioLegend), stained with Live/Dead dye (1:400; 423101 or 423113, BioLegend), FITC–anti-CD3 (1:100; 100203, BioLegend), APC–anti-CD8a (1:100; 100712, BioLegend) antibodies, and PE-conjugated OVA-tetramer or Adpgk-tetramer (1:200; HG08T14028, Helixgen). All stained cells were examined by flow cytometry. The data were collected using BD FACSDiva software and analysed by FlowJo (v10.8).
Tumour models
For therapeutic vaccine experiments, all peptides, diABZI and SABER–peptide conjugates were encapsulated in LNPs and used at 10 nmol per mouse, except for poly-I:C, which was used as carrier-free at 10 μg per mouse. B16-OVA, E.G7-OVA or MC38 tumour cells were subcutaneously inoculated at 2 × 105, 5 × 105 and 1 × 105 cells per mouse, respectively. The therapeutic vaccines were subcutaneously injected at distant sites 4, 11 and 18 days after tumour inoculation. In the B16F10 therapeutic vaccine model, tumour cells were subcutaneously inoculated at 1.5 × 105 cells per mouse. Peptides with various adjuvants or SABER–peptide conjugates were subcutaneously injected at distant sites 6, 13 and 20 days after tumour inoculation, in the presence or absence of 200 μg anti-PD1 (BE0146, BioXcell) injected intraperitoneally 1 and 4 days after each immunization. For the B16-OVA prophylactic vaccine model, 2 × 105 tumour cells were subcutaneously injected 7 days after the last immunization. Tumours were monitored by digital callipers, and the tumour volumes were calculated as 1/2 × L × W2, in which L and W are the long and short diameters of the tumours, respectively. When the tumour size exceeded 1,500 mm3 or the longest diameter exceeded 1.5 cm, the last data were recorded and mice were then euthanized. Tumour sizes were maintained in compliance with Institutional Animal Care and Use Committee guidelines (up to 2 cm in one dimension) in any of the experiments. For the B16-OVA lung metastasis model, mice were injected intravenously with 2 × 105 tumour cells 14 days after the last immunization. Mice were euthanized 18 days after tumour injection, and metastatic foci of the lungs were counted. For MC38 tumour rechallenge, 1 × 105 MC38 cells were subcutaneously injected at distant sites in mice with complete tumour regression 90 days after the primary tumour inoculation. Naive mice without previous tumour inoculation were challenged with 1 × 105 MC38 cells as controls. To analyse MC38-specific memory T cells, mice were euthanized 60 days after rechallenge. Splenocytes were blocked by anti-CD16/CD32 (1:400; 101301, BioLegend), stained with Live/Dead dye (1:400; 423107, BioLegend), FITC–anti-CD3 (1:100; 100203, BioLegend), APC–anti-CD8a (1:100; 100712, BioLegend), PB–anti-CD44 (1:100; 156005, BioLegend), APC/Cy7–anti-CD62L (1:100; 104427, BioLegend), BV650–anti-CD127 (1:100; 135043, BioLegend) and PE-conjugated Adpgk-tetramer (1:200; HG08T23002, Helixgen). All stained cells were examined by flow cytometry. The data were collected using BD FACSDiva software and analysed by FlowJo (v10.8).
Biodistribution of LNPs
B16-OVA tumour cells were subcutaneously inoculated into C57BL/6 mice. When tumour size achieved 200 mm3, DiD-labelled LNP-encapsulated AMB5–OVA was subcutaneously injected at 10 nmol per mouse. The mice were euthanized 24 h later. The heart, liver, spleen, lungs, kidneys, drain lymph node and tumour were collected and imaged by an imaging system (IVIS Lumina III, PerkinElmer). To identify DC subtypes taking DiD-labelled LNPs, draining lymph nodes were collected 24 h later. Single cells were obtained by passing the tissues through 40-μm cell strainers, lysed with ACK lysis buffer, blocked by anti-CD16/CD32 (1:400; 101301, BioLegend), and stained with FITC–anti-CD11c (1:100; 117306, BioLegend), BV711–anti-CD8a (1:100; 100747, BioLegend), APC/Cy7–anti-CD103 (1:100; 121431, BioLegend), PE–anti-CD11b (1:100; 101207, BioLegend), PE/Cy7–anti-H2-Kb(SIINFEKL) (1:100; 141607, BioLegend) and PB–anti-CD86 (1:100; 105021, BioLegend). The data were collected using BD FACSDiva software and analysed by FlowJo (v10.8).
Injection site local reactions, blood chemistry and cytokines
C57BL/6 mice were subcutaneously immunized with 10 nmol of LNP-encapsulated ABM5–OVA or diABZI + OVA. The injection sites were photographed at the indicated time. Serum was collected on days 0, 1, 4 and 8. Mouse IFNβ and TNF were measured by ELISA kits (439404 and 430904, BioLegend). Blood chemistry, including alanine transaminase (ALT), aspartate transferase (AST), creatinine and urea levels, was examined on days 0, 8 and 14 by an auto-chemistry analyzer and associated kits (BS-240VET, Mindray). Normal ranges of blood chemistry in C57BL/6 mice were delineated according to the manufacturer’s manual in the kit.
SARS-CoV-2 vaccination and challenge
In the experiments using ABM5–SNT as a standalone vaccine, C57BL/6 or hACE2-Tg mice were subcutaneously immunized with 30 nmol of LNP-encapsulated ABM5–SNT or diABZI + SNT on days 0 and 14. In the experiments using ABM5–SNT as an adjuvant for subunit vaccines, C57BL/6 mice were subcutaneously immunized with 10 μg RBD-Fc or 100 μg OVA, in the presence of 10 nmol LNP-encapsulated ABM5–SNT or diABZI + SNT on days 0 and 14. Spleens and sera were collected on day 21. The intracellular cytokine assay for SARS-CoV-2 N protein antigens was performed according to a previous study54. In brief, splenocytes were stimulated with 8 μg ml−1 SNT or 2 μg ml−1 SARS-CoV-2 spike peptide pool (PP003, Sino Biological) for 6 h in the presence of 4 μg ml−1 anti-CD28 (1:400; 102102, BioLegend) and Cell Stimulation Cocktail (1:500; 00-4970-93, Invitrogen). Brefeldin A (10 μg ml−1; S1536, Beyotime) was added to the cell culture 4 h before harvest. Cells treated with anti-CD28, cocktail and brefeldin A but without peptide stimulation served as a negative control. Cells were stained with Live/Dead dye (1:400; 423101, BioLegend), blocked by anti-CD16/CD32 (1:400; 101301, BioLegend) and stained with antibodies against T cell markers. Then, cells were fixed and permeabilized by Fix/Perm solution (554714, BD), washed with Perm/Wash buffer (554723, BD) and stained by antibodies to cytokines. For CD8+ T cells, FITC–anti-CD3 (1:100; 100203, BioLegend), PE–anti-CD8b.2 (1:100; 140408, BioLegend) and APC–anti-IFNγ (1:100; 505810, BioLegend) were used. For CD4+ T cells, FITC–anti-CD3 (1:100; 100203, BioLegend), PE–anti-CD4 (1:100; 100408, BioLegend), APC–anti-IFNγ (1:100; 505810, BioLegend) and PE/Cy7–anti-IL-4 (1:100; 504117, BioLegend) were used. All stained cells were examined by flow cytometry. The data were collected using BD FACSDiva software and analysed by FlowJo (v10.8).
The SARS-CoV-2 challenge experiment was conducted as previously reported38,55. The vaccinated hACE2-Tg mice were anaesthetized and challenged with intranasal instillation of 5,000 plaque-forming unit SARS-CoV-2 Omicron BA.5.2 variant. The behaviour and clinical signs of the mice were recorded daily after infection. Mice were euthanized to dissect lung and brain tissues for viral load determination 4 days after the challenge. Tissues were homogenized in TRIzol (15596026, Thermo Fisher) using an electric homogenizer. Total RNA was extracted and the SARS-CoV-2 subgenomic E gene was quantified by RT–qPCR using specific primers and probes: SGMRNA-E-F: CGATCTCTTGTAGATCTGTTCTC; SGMRNA-E-R: ATATTGCAGCAGTACGCACACA; and SGMRNA-E-probe: FAM-ACACTAGCCATCCTTACTGC GCTTCG-BHQ1. The SARS-CoV-2 E gene was cloned into the pcDNA3.1 vector and a standard plasmid was constructed. The viral load in the samples was determined and calculated from the standard curves.
ELISpot
Splenocytes were stimulated with 8 μg ml−1 SNT peptides. Mouse IFNγ precoated ELISpot kit (DKW22-2000-096, Dakewe) was used for the analysis of IFNγ production on hydrophobic PVDF. Spots were counted by an ELISpot reader (S6 Ultra, C.T.L.).
ELISA
An indirect ELISA was performed to determine antibody titres against OVA or WT RBD. A 96-well plate was coated with 1 μg of OVA or 0.1 μg RBD protein per well, blocked with 5% BSA and washed by PBST (0.05% Tween-20 in PBS). Mouse serum was diluted in PBST and incubated with the coated plates. After washing, HRP-conjugated anti-mouse IgG (1:3,000; 7076, Cell Signaling Technology) was incubated as secondary antibodies. After extensive washing, antigen-specific IgGs were quantified by using 3,3′,5,5′-tetramethylbenzidine (P0209, Beyotime) and measuring OD450 by a microplate reader (Varioskan Lux, Thermo Fisher). End point titres were determined as the highest reciprocal serum dilution with optical density at 450 nm (OD450) value 2.1-fold over the background.
Pseudovirus neutralization assay
Generation of pseudoviruses and the neutralization assay were performed as previously described56,57. In brief, pseudoviruses expressing S protein of WT, Delta, BA.1 or BA.5 strains were obtained by transfection of pcDNA3.1-spike (WT, Delta, BA.1 or BA.5), psPAX2 and pLenti-CMV-Puro-Luc (168w-1) plasmids into HEK293T. The culture supernatants containing luciferase-expressing pseudoviruses were harvested, titrated and stored at −80 °C until use. The hACE2-293T cells (2 × 104 cells per well) were seeded in the black flat-bottom 96-well plates. Heat-inactivated sera were first diluted 25-fold then 4-fold serially diluted in DMEM, and subsequently co-incubated with the same volume of pseudoviruses for 1 h at 37 °C. The mixtures were supplied with 10 μg ml−1 polybrene (C0351, Beyotime) to hACE2-293T cells for 6-h absorption. The mixtures were discarded, and fresh culture medium was added and incubated for an additional 42 h. After incubation, the infected cells were lysed using firefly luciferase lysis buffer (RG126M, Beyotime) and incubated with the luciferase substrate (RG058M, Beyotime). The relative light unit was measured using a microplate reader. The half pseudovirus neutralization titres were determined with a four-parameter non-linear regression curve by GraphPad Prism (v9.5.1).
Statistical analysis
All statistical analyses were performed using GraphPad Prism (v9.5.1). Data were shown as mean ± s.e.m., geometric mean ± 95% CI, violin plot, or box and whiskers as indicated in each experiment. Comparisons of two groups were assessed using two-tailed Student’s t-test. Comparisons of multiple groups were assessed using one-way ANOVA with Tukey’s multiple comparisons test, Kruskal–Wallis with Dunn’s multiple comparisons test or two-way ANOVA. Survival curves were compared using the log-rank test. The investigators were not blinded to the experiments that were carried out under highly standardized and predefined conditions, except for microscopy images and tumour measurement, which were evaluated in an investigator-blind manner.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
All data supporting the findings of this study are available within the paper and Supplementary Information. Experimental structures used from the Protein Data Bank are under the accession numbers 4F9G (ref. 15), 6DXL (ref. 16), 6XNP (ref. 17) and 4KSY (ref. 18). Source data are provided with this paper.
References
Lin, M. J. et al. Cancer vaccines: the next immunotherapy frontier. Nat. Cancer 3, 911–926 (2022).
Wang, J. et al. Pulmonary surfactant-biomimetic nanoparticles potentiate heterosubtypic influenza immunity. Science 367, eaau0810 (2020).
Wherry, E. J. & Barouch, D. H. T cell immunity to COVID-19 vaccines. Science 377, 821–822 (2022).
Cruz, F. M., Colbert, J. D., Merino, E., Kriegsman, B. A. & Rock, K. L. The biology and underlying mechanisms of cross-presentation of exogenous antigens on MHC-I molecules. Annu. Rev. Immunol. 35, 149–176 (2017).
Sellars, M. C., Wu, C. J. & Fritsch, E. F. Cancer vaccines: building a bridge over troubled waters. Cell 185, 2770–2788 (2022).
Takeuchi, O. & Akira, S. Pattern recognition receptors and inflammation. Cell 140, 805–820 (2010).
Wang, J. et al. Physical activation of innate immunity by spiky particles. Nat. Nanotechnol. 13, 1078–1086 (2018).
Zhao, T. et al. Vaccine adjuvants: mechanisms and platforms. Signal Transduct. Target. Ther. 8, 283 (2023).
Minev, B. R., Chavez, F. L., Dudouet, B. M. & Mitchell, M. S. Synthetic insertion signal sequences enhance MHC class I presentation of a peptide from the melanoma antigen MART-1. Eur. J. Immunol. 30, 2115–2124 (2000).
Hayashi, A. et al. A strategy for efficient cross-presentation of CTL-epitope peptides leading to enhanced induction of in vivo tumor immunity. J. Control. Release 117, 11–19 (2007).
Anderson, K. et al. Endogenously synthesized peptide with an endoplasmic reticulum signal sequence sensitizes antigen processing mutant cells to class I-restricted cell-mediated lysis. J. Exp. Med. 174, 489–492 (1991).
Sneh-Edri, H., Likhtenshtein, D. & Stepensky, D. Intracellular targeting of PLGA nanoparticles encapsulating antigenic peptide to the endoplasmic reticulum of dendritic cells and its effect on antigen cross-presentation in vitro. Mol. Pharm. 8, 1266–1275 (2011).
Ishikawa, H. & Barber, G. N. STING is an endoplasmic reticulum adaptor that facilitates innate immune signalling. Nature 455, 674–678 (2008).
Wu, J. et al. Cyclic GMP-AMP is an endogenous second messenger in innate immune signaling by cytosolic DNA. Science 339, 826–830 (2013).
Yin, Q. et al. Cyclic di-GMP sensing via the innate immune signaling protein STING. Mol. Cell 46, 735–745 (2012).
Ramanjulu, J. M. et al. Design of amidobenzimidazole STING receptor agonists with systemic activity. Nature 564, 439–443 (2018).
Chin, E. N. et al. Antitumor activity of a systemic STING-activating non-nucleotide cGAMP mimetic. Science 369, 993–999 (2020).
Zhang, X. et al. Cyclic GMP-AMP containing mixed phosphodiester linkages is an endogenous high-affinity ligand for STING. Mol. Cell 51, 226–235 (2013).
Rosalia, R. A. et al. Dendritic cells process synthetic long peptides better than whole protein, improving antigen presentation and T-cell activation. Eur. J. Immunol. 43, 2554–2565 (2013).
Lahnsteiner, M. et al. Improving the stability of maleimide-thiol conjugation for drug targeting. Chemistry 26, 15867–15870 (2020).
Bijker, M. S. et al. Superior induction of anti-tumor CTL immunity by extended peptide vaccines involves prolonged, DC-focused antigen presentation. Eur. J. Immunol. 38, 1033–1042 (2008).
Wilson, D. S. et al. Antigens reversibly conjugated to a polymeric glyco-adjuvant induce protective humoral and cellular immunity. Nat. Mater. 18, 175–185 (2019).
Lee, Y. S. et al. Peptides derived from S and N proteins of severe acute respiratory syndrome coronavirus 2 induce T cell responses: a proof of concept for T cell vaccines. Front. Microbiol. 12, 732450 (2021).
Ergun, S. L., Fernandez, D., Weiss, T. M. & Li, L. STING polymer structure reveals mechanisms for activation, hyperactivation, and inhibition. Cell 178, 290–301.e10 (2019).
Zhang, C. et al. Structural basis of STING binding with and phosphorylation by TBK1. Nature 567, 394–398 (2019).
Yu, X. et al. The STING phase-separator suppresses innate immune signalling. Nat. Cell Biol. 23, 330–340 (2021).
Ablasser, A. et al. Cell intrinsic immunity spreads to bystander cells via the intercellular transfer of cGAMP. Nature 503, 530–534 (2013).
Hou, X., Zaks, T., Langer, R. & Dong, Y. Lipid nanoparticles for mRNA delivery. Nat. Rev. Mater. 6, 1078–1094 (2021).
Kalies, K. U., Allan, S., Sergeyenko, T., Kröger, H. & Römisch, K. The protein translocation channel binds proteasomes to the endoplasmic reticulum membrane. EMBO J. 24, 2284–2293 (2005).
Haag, S. M. et al. Targeting STING with covalent small-molecule inhibitors. Nature 559, 269–273 (2018).
Guo, C. et al. SLC38A2 and glutamine signalling in cDC1s dictate anti-tumour immunity. Nature 620, 200–208 (2023).
Chu, T. T. et al. Tonic prime-boost of STING signalling mediates Niemann-Pick disease type C. Nature 596, 570–575 (2021).
Liu, B. et al. Human STING is a proton channel. Science 381, 508–514 (2023).
Yadav, M. et al. Predicting immunogenic tumour mutations by combining mass spectrometry and exome sequencing. Nature 515, 572–576 (2014).
Kreiter, S. et al. Mutant MHC class II epitopes drive therapeutic immune responses to cancer. Nature 520, 692–696 (2015).
Ott, P. A. et al. An immunogenic personal neoantigen vaccine for patients with melanoma. Nature 547, 217–221 (2017).
Nelde, A. et al. SARS-CoV-2-derived peptides define heterologous and COVID-19-induced T cell recognition. Nat. Immunol. 22, 74–85 (2021).
Liu, Z. et al. A novel STING agonist-adjuvanted pan-sarbecovirus vaccine elicits potent and durable neutralizing antibody and T cell responses in mice, rabbits and NHPs. Cell Res. 32, 269–287 (2022).
Hilf, N. et al. Actively personalized vaccination trial for newly diagnosed glioblastoma. Nature 565, 240–245 (2019).
Awad, M. M. et al. Personalized neoantigen vaccine NEO-PV-01 with chemotherapy and anti-PD-1 as first-line treatment for non-squamous non-small cell lung cancer. Cancer Cell 40, 1010–1026.e11 (2022).
Pleguezuelos, O. et al. Efficacy of FLU-v, a broad-spectrum influenza vaccine, in a randomized phase IIb human influenza challenge study. NPJ Vaccines 5, 22 (2020).
Heitmann, J. S. et al. A COVID-19 peptide vaccine for the induction of SARS-CoV-2 T cell immunity. Nature 601, 617–622 (2022).
Liu, H. et al. Structure-based programming of lymph-node targeting in molecular vaccines. Nature 507, 519–522 (2014).
Schudel, A. et al. Programmable multistage drug delivery to lymph nodes. Nat. Nanotechnol. 15, 491–499 (2020).
Wang, H. et al. Metabolic labeling and targeted modulation of dendritic cells. Nat. Mater. 19, 1244–1252 (2020).
Kranz, L. M. et al. Systemic RNA delivery to dendritic cells exploits antiviral defence for cancer immunotherapy. Nature 534, 396–401 (2016).
Jackson, S. et al. Immunogenicity of a two-dose investigational hepatitis B vaccine, HBsAg-1018, using a Toll-like receptor 9 agonist adjuvant compared with a licensed hepatitis B vaccine in adults. Vaccine 36, 668–674 (2018).
Aung, A. et al. Low protease activity in B cell follicles promotes retention of intact antigens after immunization. Science 379, eabn8934 (2023).
Mayer, C. T. et al. Selective and efficient generation of functional Batf3-dependent CD103+ dendritic cells from mouse bone marrow. Blood 124, 3081–3091 (2014).
Koerner, J. et al. PLGA-particle vaccine carrying TLR3/RIG-I ligand riboxxim synergizes with immune checkpoint blockade for effective anti-cancer immunotherapy. Nat. Commun. 12, 2935 (2021).
Barbet, G. et al. TAP dysfunction in dendritic cells enables noncanonical cross-presentation for T cell priming. Nat. Immunol. 22, 497–509 (2021).
Hao, H. et al. Immunization against Zika by entrapping live virus in a subcutaneous self-adjuvanting hydrogel. Nat. Biomed. Eng. 7, 928–942 (2023).
Lendemans, D. G., Myschik, J., Hook, S. & Rades, T. Immuno-stimulating complexes prepared by ethanol injection. J. Pharm. Pharmacol. 57, 729–733 (2005).
Li, B. et al. A conserved N protein nano-vaccine of COVID-19 exerts potent and cross-reactive humoral and cellular immune responses in mice. J. Med. Virol. 95, e29115 (2023).
Tada, T., Peng, J. Y., Dcosta, B. M. & Landau, N. R. Single-epitope T cell-based vaccine protects against SARS-CoV-2 infection in a preclinical animal model. JCI Insight https://doi.org/10.1172/jci.insight.167306 (2023).
Wang, X. et al. An engineered recombinant protein containing three structural domains in SARS-CoV-2 S2 protein has potential to act as a pan-human coronavirus entry inhibitor or vaccine antigen. Emerg. Microbes Infect. 12, 2244084 (2023).
Wang, J. et al. A fourth dose of the inactivated SARS-CoV-2 vaccine redistributes humoral immunity to the N-terminal domain. Nat. Commun. 13, 6866 (2022).
Acknowledgements
We thank C. Y. Yang at Sun Yat-sen University for the OT-I mice; Z. Jiang at Peking University and H. Yin at Tsinghua University for the STING−/− mice and the STING−/− HeLa cells; T. Chu at the Shenzhen Bay Laboratory for the STING–APEX2 plasmid; and Q. Wang at FDU SHMC and the core facility of IPM FAH SYSU for technical support. Schematic diagrams were created using BioRender (https://biorender.com). The work is supported by the National Key Research and Development Program of China (2022YFC2305800 and 2023YFC2307800), the National Natural Science Foundation of China (82341042, 82425033, T2225010, 22007105, 82341036, 32270993, 32300740 and 92369112) and the Liaoning Provincial Department of Education Scientific Research Project (LJKMZ20220455).
Author information
Authors and Affiliations
Contributions
J.W. conceived the SABER concept and provided overall direction. J.W., L.L. and Y.Z. co-supervised the study. X.W. identified and synthesized all compounds, and designed and performed most of the in vitro experiments. Z.H. designed and performed most of the in vivo experiments. L.X. performed the SARS-CoV-2 vaccination and challenge experiments. L.S. performed ER isolation and TEM, and helped with tumour models. J.J. performed the OT-I and STING–APEX2 assays. C.D. helped with the DC culture and neutralization assay. W.Y., L.P. and Hao Yang helped with RT–qPCR, ELISA and in vivo experiments. X.Z., X.L., Haolan Yang, Y.C., Y.L., J.L. and W.X. helped with in vitro and in vivo experiments. W.L. and X.Xia help with the transgenic mice study. Z.L. contributed RBD-Fc and helped with the SARS-CoV-2 vaccination. X.W., Z.H., L.S., Y.Z. and J.W. performed the statistical analysis and diagram drawing. J.W., Y.Z., X.W. and Z.H. drafted the manuscript. L.L., S.J., W.L. and X.Xie made a critical revision.
Corresponding authors
Ethics declarations
Competing interests
J.W., X.W., Z.H., L.S., J.J. and C.D. are inventors on a patent related to the ER-targeting technology (ZL202111251211.9). All other authors declare no competing interests.
Peer review
Peer review information
Nature thanks Ed Lavelle, David Sancho and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
Extended Data Fig. 1 Synthesis of the SABER family.
a-d, Crystal structures of STING and its agonists. A “closed lid” conformation was induced by 2’3’-cGAMP (a, PDB: 4KSY) and SR717 (b, PDB: 6XNP), whereas an “open lid” conformation was induced by c-diGMP (c, PDB: 4F9G) and compound 2 (d, PDB: 6DXL). e, The interaction diagram of compound 2 and STING (PDB: 6DXL) was analysed by Ligplot+ software. f, Synthesis route. Reactions and conditions: i) aq. NH4OH, 24 h, then 50 °C, 2 h; ii) N, N-Diisopropylethylamine (DIPEA), N, N-Dimethylformamide (DMF), 120 °C, 16 h; iii) Trifluoroacetic acid (TFA), Dichloromethane (DCM), 2 h; iv) BBr3, DCM, 16 h; v) DIPEA, DMF, 80 °C, 5 h; vi) DIPEA, isopropanol, 120 °C, 16 h; vii) NaS2O4, MeOH, 25 min; viii) 1-ethyl-3-methyl-1H-pyrazole-5-carbonyl isothiocyanate, DMF, 30 min then n-(3-dimethylamin opropyl)-n’-ethylcarbodiimide hydrochloride (EDCI), DIPEA, 14 h; ix) TFA, DCM, 2 h, then a) DIPEA, DMF, 1 h or b) EDCI, 1-Hydroxyben zotriazole (HOBt), DIPEA, DMF, 2 h or c) DIPEA, DMF, overnight, then TFA, DCM, 2 h then DIPEA, DMF, 1 h.
Extended Data Fig. 2 SABER conjugated peptides activate STING.
a, Mechanism of intro-cyclization reaction of ABM5-OVA between the maleimide and C-terminal cysteine. b, c, The cyclization reaction inside ABM5-OVA in the addition of DIPEA was confirmed by high-performance liquid chromatography (HPLC) and matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF). d, The chemical structure of SR717-OVA. EC50 of ABM5-OVA or SR717-OVA was determined by measuring IFN-β expression in BMDCs for 24 h by ELISA. e, BMDCs were treated with 1 μM diABZI or ABM5-OVA for 1-10 h. Phosphorylation of STING and TBK1 was measured by Western blot. For gel source data, see Supplementary Fig. 1b. f, BMDCs were incubated with 50 nM of OVA peptide, diABZI + OVA, or ABM5-OVA for 8 h. The MFI of H2-Kb (SIINFEKL) complex was measured by flow cytometry. n = 3 biologically independent samples. g, h, Peptides containing a mutated weak OVA CD8+ T cell epitope SIINFEKI (OVA-i) was used to validate the capability of SABER. BMDCs were incubated with 0.5 μM of OVA-i, diABZI + OVA-i, or ABM5-OVA-i for 8 h. The MFI of H2-Kb (SIINFEKL) complex (g) and CD86 (h) were analysed by flow cytometry. n = 3 biologically independent samples. Data are mean ± s.e.m. One-way ANOVA with Tukey’s multiple comparisons test for f-h. Data are representative of three independent experiments.
Extended Data Fig. 3 Cleavage of ABD-S-OVA.
a, ABD, a modified SABER molecule containing a disulfide bond, was designed and synthesized. b, Schematic diagram of cleavage of ABD-S-OVA. ABD-S-OVA could be cleaved by GSH in the cytosol, while ABM5-OVA could not. The EC50 of ABD-S-OVA was measured as Fig. 1a. The schematic was created using BioRender (https://biorender.com). c, Cleavage of ABM5-OVA and ABD-S-OVA in PBS or PBS containing 1 or 10 mM GSH at 37 °C was analysed by HPLC. d-h, Cleavage products were confirmed by MS. Data are representative of three independent experiments.
Extended Data Fig. 4 Mechanism of enhanced cross-presentation.
a, HeLa cells were incubated with OVA-Cy5.5 or ABM5-OVA-Cy5.5 for 8 h and analysed by flow cytometry. n = 3 biologically independent samples. b, HeLa cells were incubated with 500 nM ABM5-OVA-Cy5.5 or ABM5-SNT-Cy5.5 for 2 h, washed and harvested 2 h later. n = 4 biologically independent samples. c, HeLa cells were incubated with 5 μM ABM5-OVA or ABM5-SNT for 2 h, washed and harvested 4 h later. Cxcl10 was measured by RT-qPCR. n = 3 biologically independent samples. d, e, STING-/- HeLa cells expressing STING-flag were incubated with 2.5 μM free ABM5- OVA/SNT-Cy5.5, or 1 μM LNP-encapsulated ones for 2 h, stained and analysed by confocal microscopy. Scale bar, 5 μm. The peptide-STING colocalization was summarized in (e). n = 10 cells. f, g, BMDCs were pre-treated with 10 μM diABZI or imiquimod for 40 min, and then incubated with 0.5 μM ABM5-OVA for 8 h. The MFI of H2-Kb (SIINFEKL) complex (f) and CD86 (g) were analysed. n = 3 biologically independent samples. h, i, BMDCs were incubated with 0.5 μM of ABM5-OVA-i or ABD-S-OVA-i for 8 h. The antigen cross-presentation (h) and CD86 (i) were analysed. n = 3 biologically independent samples. j, k, STING-/- HeLa cells expressing STING-flag were incubated with 2.5 μM of diABZI + OVA-Cy5.5, ABM5-OVA-Cy5.5 or Cy5.5-OVA-ABM5 for 1-8 h. STING and peptides were visualized. Scale bar, 5 μm. The peptide-STING colocalization at 1-8 h after incubation was summarized (k). n = 10 cells. l, Schematic diagram of the cytosolic pathway of antigen cross-presentation and inhibitors. The schematic was created using BioRender (https://biorender.com). m, BMDCs were pre-treated with various inhibitors for 40 min, followed by incubation with 0.5 μM of ABM5-OVA for 8 h, and analysed for cross-presentation. n = 3 biologically independent samples. n, BMDCs were incubated with 0.5 μM of OVA-ABM5 or ABM5-OVA for 8 h, and analysed for cross-presentation. n = 3 biologically independent samples. o, BMDCs were pre-treated with 10 μM H151 for 40 min, followed by incubation with 0.5 μM of ABM5-OVA for 8 h. The CD86 was analysed. n = 3 biologically independent samples. Data are mean ± s.e.m. One-way ANOVA with Tukey’s multiple comparisons test for b, c, e-i, m-o, two-way ANOVA for k. Data are representative of three independent experiments.
Extended Data Fig. 5 SABER condenses key machinery in cross-presentation to form microreactors.
a, HeLa cells expressing STING-flag were incubated with 5 μM diABZI + OVA-biotin, or ABM5-OVA-biotin for 2 h. ER, TAP1, and proteasome (Protea) were stained by primary antibodies and STAR-Red labelled secondary antibodies, while STAR-Orange labelled secondary antibodies were used for STING. Cells were visualized by STED. The fluorescence intensity of STING and STAR-RED at the indicated region was analysed by ZEN software. Scale bar, 5 μm or 1 μm (zoom). b, Schematic diagram of cross-presentation microreactors. SABER condenses antigens, STING, TAP and proteasomes into microreactors. The schematic was created using BioRender (https://biorender.com). c, HEK293T expressing STING-APEX2 were first treated with 0.5 μM diBAZI, OVA peptide, or ABM5-OVA for 2 h, then with biotin-phenol and H2O2. Biotinylated proteins were pulled down by streptavidin magnetic beads and analyzed by western blot for TAP1 and proteasomes. STING-APEX2 and GAPDH expression in input cell lysates served as internal controls. For gel source data, see Supplementary Fig. 1c. d-i, BMDCs were incubated with 2.5 μM OVA-Cy5.5 or ABM5-OVA-Cy5.5 for 2 h. The co-localization of peptides with STING, ER (d), TAP1 (e), or proteasomes (f) was visualized by confocal microscopy. Scale bar, 10 μm. Co-localization coefficiency of the peptide with STING, ER (g), TAP1 (h), or proteasomes (i) was summarized. n = 5 cells. j, BMDCs were incubated with 2.5 µM diABZI or ABM5-OVA for 2 h, and examined by TEM. The rough endoplasmic reticulum (RER), smooth endoplasmic reticulum (SER), or the Golgi apparatus (G) were indicated. Scale bar, 1 μm. Data are mean ± s.e.m. One-way ANOVA with Tukey’s multiple comparisons test for g-i. Data are representative of three independent experiments.
Extended Data Fig. 6 Fabrication of LNPs encapsulating SABER-decorated antigens.
a, SABER-decorated antigens were encapsulated in LNPs by microfluidic mixing. The schematic was created using BioRender (https://biorender.com). b, Particle sizes and PDIs of LNPs with different lipid compositions were compared. n = 3 biologically independent samples. c, BMDCs were incubated with 100 nM ABM5-OVA encapsulated within LNPs containing different proportions of ionizable lipid SM102 or ALC0315 for 4 h, and analysed for cross-presentation by flow cytometry. n = 3 biologically independent samples. d, e, Particle sizes (d) and PDIs (e) of LNPs fabricated in this study were summarized. n = 3 biologically independent samples. f, g, LNPs were stored at 4 °C for 140 days and monitored for changes in particle sizes and PDIs. n = 3 biologically independent samples. h, EC50 of LNP-encapsulated diABZI, ABM5-OVA, and ABD-S-OVA was measured as Fig. 1a. i, THP-1-derived DC-like cells were generated by culturing THP-1 with human GM-CSF and IL-4. THP-1 or DC-like cells were incubated with carrier-free or LNP-encapsulated diABZI or ABM5-OVA. The upregulation of Cxcl10 was measured by RT-qPCR at different time points after incubation. n = 3 biologically independent samples. j-q, C57BL/6 mice were immunized with 10 nmol of LNP-encapsulated diABZI + OVA or ABM5-OVA. Local reaction of injection sites as indicated by white arrows (j). Body weight change (k), blood alanine aminotransferase (ALT) (l), aspartate aminotransferase (AST) (m), creatinine (n), urea levels (o), serum IFN-β (p) and TNF levels (q) were measured at the indicated time after injection. Dashed lines are normal ranges of blood chemistry in mice. n = 4 mice. Data are mean ± s.e.m. One-way ANOVA with Tukey’s multiple comparisons test for c. Data are representative of two independent experiments.
Extended Data Fig. 7 Immunogenicity and biodistribution of SABER-decorated antigens.
a, Mice were immunized with 10 nmol of diABZI + OVA or ABM5-OVA encapsulated in SM102-containing LNPs on days 0 and 14. OVA-tetramer+ CD8+ T cells in PBMCs were analysed on day 21. Left-right, n = 4, 5, 8 mice. b, Mice were similarly immunized with diABZI + OVA encapsulated within one particle (diABZI:OVA = 1.2), designated as LNP (diABZI + OVA), or diABZI + OVA encapsulated in different particles, or LNP-encapsulated ABM5-OVA for two doses. OVA peptides were used at 10 nmol and diABZI was kept in the same amount in all groups, except for LNP (diABZI + OVA), in which diABZI is 12 nmol. OVA-tetramer+ CD8+ T cells in PBMCs were analyzed on day 21. n = 5 mice. c, Mice were immunized with 10 nmol of carrier-free or LNP-encapsulated diABZI + OVA, ABM5-OVA, or LNP-encapsulated ABM5-Adpgk + OVA for two doses. n = 5 mice. d, e, B16-OVA tumour-bearing mice were s.c. injected with 10 nmol DiD-labelled LNP-encapsulated ABM5-OVA. The biodistribution of LNPs was measured 24 h later by IVIS. The relative fluorescence intensity was summarized in (e). n = 4 mice. f, Mice were s.c. injected with 10 nmol DiD-labelled LNP-encapsulated ABM5-OVA. Subtypes of dendritic cells taking LNPs in draining lymph nodes were analysed by flow cytometry 24 h later. n = 3 mice. g, h, Mice were immunized with 10 nmol LNP-encapsulated ABM5-OVA. H2-Kb (SIINFEKL) complex (g), and CD86 (h) on the surface of various DC subtypes were analysed 24 h later. n = 3 mice. i, Schematic diagram of ABM5-OVA30aa. j-l, Mice were immunized with 5 nmol LNP-encapsulated diABZI + OVA30aa or ABM5-OVA30aa on day 0, 14. OVA-tetramer+ CD8+ T cells in PBMCs were analyzed on day 21 (j). Mice were challenged with 2 × 105 B16-OVA tumour cells on day 21. Tumour volume (k) and survival rates (l) of B16-OVA tumour-bearing mice were monitored for 5 weeks. n = 4 mice. Data are mean ± s.e.m. One-way ANOVA with Tukey’s multiple comparisons test for a-c, e, f, j, two-way ANOVA for g, h, k, and the log-rank test for l. Data are representative of two independent experiments.
Extended Data Fig. 8 SABER-decorated peptide vaccines for tumours and SARS-CoV-2.
a, Representative flow plots of Adpgk -tetramer+ CD8+ T cells for Fig. 4a. b, Survival rates for Fig. 4d. n = 9 mice. c, The proportions of naive T cells (CD44− CD62L+), effector T cells (TEFF, CD44+ CD62L− CD127−), central memory T cells (TCM, CD44+ CD62L+) and effector memory T cells (TEM, CD44+ CD62L- CD127+) among splenic tetramer+ CD8+ T cells in Fig. 4e were analysed by flow cytometry. n = 8 mice. d, Schematic diagram of ABN2-M27 antigen. e, Schematic diagram of a peptide vaccine targeting N129-148 (SNT) in the nucleocapsid protein (N) of SARS-CoV-2. S, spike protein. M, membrane protein. E, envelope protein. f, Representative data of ELISpot in Fig. 4h. g, Representative flow plots of IFN-γ-expressing CD8+ T cells in Fig. 4i. h, Mice were immunized as Fig. 4l. OVA-specific IgG were measured 14 days after the first vaccination. n = 6 mice. i, Schematic diagram showing the adjuvant effect of ABM5-SNT vaccine on a SARS-CoV-2 RBD-Fc subunit vaccine in inducing NAbs. The schematics in panels e,i were created using BioRender (https://biorender.com). j-o, Mice were immunized as Fig. 4m. Serum NAb titres of immunized mice were measured by pseudovirus neutralizing assay for Delta (j), WT (k), BA.1 (l) and BA.5 (m). n = 6 mice. The ICC assay was further used to measure splenic IFN-γ (n) or IL-4 (o) expressing CD4+ T cells after S protein peptide pool stimulation followed by incubating with cell stimulation cocktail. n = 5 mice. Data are mean ± s.e.m for c, n, o, geometric mean ± 95% CI for h, or box and whiskers indicating median (middle line), 25-75th percentile (box), and min-max (whiskers) for j-m. One-way ANOVA with Tukey’s multiple comparisons test for c, n, o, the log-rank test for b, and Kruskal-Wallis test for h, j-m. Data are representative of two independent experiments.
Extended Data Fig. 9 Graphic summary of the study.
SABER effectively delivers antigens to ER and condenses key machinery in cross-presentation to form microreactors. Conjugation of SABER to tumour neoantigens or conserved SARS-CoV-2 antigens induces robust CD8+ T cell immune responses. SABER boasts about 7- to 150-fold superior effectiveness to a mixture of STING agonists and antigens, indicating that precise subcellular delivery targeting the “Last Mile” of cross-presentation can make a substantial difference. The graphic summary was created using BioRender (https://biorender.com).
Supplementary information
Supplementary Information
This file contains detailed chemical synthesis of compounds, Supplementary Figure 1 (Uncropped immunoblot images with ladders), Supplementary Figure 2 (Gating strategies for flow cytometry analysis)
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Wang, X., Huang, Z., Xing, L. et al. STING agonist-based ER-targeting molecules boost antigen cross-presentation. Nature 641, 202–210 (2025). https://doi.org/10.1038/s41586-025-08758-w
Received:
Accepted:
Published:
Version of record:
Issue date:
DOI: https://doi.org/10.1038/s41586-025-08758-w
This article is cited by
-
Eliciting antitumor immunity via therapeutic cancer vaccines
Cellular & Molecular Immunology (2025)
-
Integrated bioinformatics and experimental validation identify lysosome and immune infiltration-related genes as therapeutic targets in late-onset major depressive disorder
Scientific Reports (2025)
-
Crucial meeting: molecule helps vaccine to interact with killer T cells
Nature (2025)